JP2006068300A - Body condition monitoring apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a body condition monitoring apparatus which can precisely detect a movement of a user (a posture, a fall and a walking rhythm) on real time. <P>SOLUTION: This body condition monitoring apparatus includes an acceleration sensor which senses the accelerations of three orthogonal axes and measures the individual acceleration data, a sensor unit having a data processing apparatus which has an initializing function to convert the data from the acceleration sensor to a triaxial coordinate system and a mounting means to mount the sensor unit on the user's body, and is mounted on the body of the user to collect the information related to the user's movement. Since a signal from the triaxial acceleration sensor is processed by the built-in data processing apparatus, the information related to the various movements, the posture, the fall, or the like, of the user can be collected and can be utilized for the health care, the sports training, or the like. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a body condition monitoring apparatus using an acceleration sensor.

  Patent Document 1 is a portable accident monitoring device that analyzes acceleration data of three axes orthogonal to each other and transmits an abnormal signal wirelessly when a fall state for a predetermined time or more or an acceleration of a predetermined value or more is detected. Are listed. Detection of a fall state or abnormality is determined using a temporal threshold and an amplitude threshold. Further, as a means for detecting a person's location, the location is estimated based on electric field strength data from a plurality of base stations held in a PHS terminal.

  Patent Document 2 describes that a biological information collection device including a wristwatch-type display device incorporating an acceleration sensor or the like is attached to a user and the biological information is stored in a memory of the collection device. By analyzing the output of the acceleration sensor, it is possible to detect the movement of the user and determine whether it is a movement during sleep or a movement during awakening. This biological information collecting apparatus is connected to a docking station connected to the PC, and the user's mental condition is interrogated (why such movement is performed) according to a program incorporated in the PC.

In Patent Document 3, acceleration is measured on a human body with a two-axis or three-axis accelerometer, and a static state or a dynamic state is determined by a magnitude of a change in acceleration in each direction over a predetermined time. Further, it is determined whether the state is stable or dangerous based on the magnitude of change in acceleration. That is, specific dynamic states such as walking, running, stair climbing, falling (including the front, back, left and right falling directions) and specific static states such as stopping, crouching, leaning upward, leaning sideways, leaning downward Judgment is automatically made by waveform analysis based on the magnitude of change. The threshold value of acceleration change used for determination is 0.5 G, and the data interval is 5 seconds. The determination interval can be appropriately set for 30 minutes or shorter depending on each individual and the behavioral situation. In addition, when it is determined as a dangerous state, the location of the subject can be detected by PHS or GPS.
Japanese Patent Laid-Open No. 10-295649 Japanese Patent Laid-Open No. 2003-290176 JP 2004-81632 A

  Conventional biological information processing technology using an acceleration sensor cannot simultaneously detect a user's posture, falls, walking rhythm, etc. using a single device. Also, there was a lack of reliable data processing techniques for accurate detection. In addition, there are disadvantages that appropriate data processing cannot be performed depending on the sensor mounting position and that a detection error is large when walking at a low speed. The present invention is a data processing method capable of detecting an acceleration signal and a gravity vector in three orthogonal directions using one sensor device, and accurately detecting a user's movement (posture, fall, walking rhythm) in real time. It aims at providing a physical condition monitoring device.

    The monitoring apparatus according to the present invention detects an acceleration of three orthogonal axes and measures each acceleration data, and data processing having an initialization function for converting data from the acceleration sensor into a three-axis coordinate system A sensor unit having the device and an attachment means for attaching the sensor unit to the user's body are attached to the user's body to collect information on the user's movement.

  The attachment means is a means for fixing the monitoring device to the user's body so that there is no relative movement between the body and the body, such as a button of clothes, a belt buckle, a brooch, a brassiere, a hot body warmer, etc. It can be attached to sticky pads.

  Specifically, the data processing device provides a processor for processing a detection output from the acceleration sensor, a ROM (read only memory) for storing a computer program executed by the processor, and a work area for the processor. A RAM (random access memory) is provided for temporarily storing the results of processing by the processor.

  According to the present invention, since signals from the three-axis acceleration sensor are processed by the built-in data processing device, it is possible to collect information on various movements, postures, falls, etc. of the user, and health management and sports training It can be used for

  Furthermore, in one form of this invention, the monitoring apparatus is comprised so that a user's fall may be detected based on the DC signal component of the z-axis (vertical direction) from an acceleration sensor. The direct current component of the signal from the acceleration sensor represents the direction of gravity of the user. A slow fluctuation component represents a relatively slow movement. When the user falls, if the DC component in the vertical direction of the acceleration sensor is 1 before the fall, the DC component becomes 0 because the direction of gravity has changed after the fall. Therefore, the user's fall can be detected based on the direct current component in the vertical direction.

  In one embodiment of the present invention, the monitoring device indicates the direction of the user's fall based on the DC component of the signal of the front-rear axis (y-axis) or the left-right axis (x-axis) from the triaxial acceleration sensor. Configured to detect. The y-axis signal represents the user's longitudinal movement. If the DC component of the y-axis is 0 before the user falls forward, the value becomes 1 after the fall. When the user falls backward, the y-axis DC component is -1. Therefore, it is possible to detect the front-rear direction of the fall based on the DC component of the y-axis. Similarly, the left-right direction of the user falling can be detected based on the DC component of the x axis.

  In one form of this invention, the monitoring device is configured to detect a user's fall when the difference in the moving average value of the DC component of the signal in the vertical direction is greater than a predetermined value. That is, in order to detect a change in the DC component, a difference between moving average values over a certain period of time is taken, and when this value is greater than a predetermined value, a fall is determined.

  In a further aspect of the present invention, the monitoring device is configured such that the difference between the moving average values of the DC components of the signals of the front-rear direction axis (y-axis) or the left-right direction axis (x-axis) from the triaxial acceleration sensor is a predetermined value. When larger, the direction of the user's fall is detected. For the same purpose as the signal processing on the vertical axis, the difference between the moving average values over a certain period of time is taken for the DC component of the y axis and the x axis, and when this value is greater than a predetermined value, the direction of the fall is determined. .

  Moreover, in one form of this invention, the monitoring apparatus is comprised so that a user's attitude | position may be detected based on the moving average value of the fluctuation component of the signal of each of the three axes from the three-axis acceleration sensor. The fluctuation component of the signal from the acceleration sensor represents a change in speed (acceleration). Therefore, the composite vector of the three-axis signal changes, and by detecting this composite vector, the user's posture (standing, sleeping on the side, sleeping on the bottom, sleeping on the top, etc.) Can be determined.

  In an embodiment of the present invention, the monitoring device includes a communication unit that communicates with a user's mobile phone or a mobile terminal device, and when the abnormality is detected by the user as a result of the processing by the processor, the monitoring unit is provided. It is configured to start up and transmit data to the computer of the information processing center via the mobile phone or the mobile terminal device.

  Furthermore, in one aspect of the present invention, the computer of the information processing center processes the data in real time to determine the user's situation in response to the reception of data related to the user's fall sent from the monitoring device. , Configured to send an advisory message to the user or establish a voice link between the supporter and the user.

  Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows a fuselage region (shaded portion) to which a sensor according to an embodiment of the present invention can be attached and a definition of a three-dimensional coordinate system (actual coordinate system Cr and reference coordinate system Cs). The monitoring device 10 is a sensor unit 14 that can be attached to a user's clothing buttons, belt buckles, brooches, bras, sticky pads such as hot warmers, etc., and sensor unit 14 coupled with Bluetooth or infrared Mobile phone 16 for communication. A mobile terminal device (PDA) can be used instead of the mobile phone 16.

  FIG. 2 is a diagram for defining an actual coordinate system at the time of mounting the sensor and a reference coordinate system at the time of calculation. The three-dimensional coordinate system (actual coordinate system Cr) at the time of actual mounting does not necessarily coincide with the reference coordinate system Cs for calculation. Since it is possible to convert from an arbitrary Cr coordinate system to a Cs coordinate system using the affine transformation matrix obtained in the initialization process of FIG. 6 described later, it is not necessary to pay attention to the axis alignment at the time of wearing. The transformation of the three-dimensional space has six degrees of freedom, and if there is no movement of the origin, there are three degrees of freedom (α, β, γ). A specific method will be described later with reference to FIG.

  FIG. 3 shows the arrangement of the sensor unit 14 and the two biaxial acceleration sensors 15-1 and 15-2. The main components of the sensor unit 14 include a CPU 19, a Bluetooth chip 13, a wireless data communication antenna 11 and a battery 17 in addition to the acceleration IC 15.

  As shown in FIG. 4, the sensor unit 14 can be attached to, for example, a broach (a), a belt buckle (b), a brassiere (c), or the like by means of a magic tape, hook or snap. It can also be attached to a part of clothes, for example, a button, or pasted like a hot body warmer.

  Acceleration ICs 15-1 and 15-2 are sensors using film-like piezoelectric elements, such as the 2-axis acceleration sensor ADXL202 (trade name) from Analog Devices, Inc., or the 3-axis acceleration sensor ACH-04 from Tokyo Sensor Co., Ltd. Product name) or the like. In this embodiment, one biaxial acceleration IC 15-1 is arranged with the lateral direction of the body as the x axis and the vertical direction as the z axis, and the other two axis acceleration IC15-2 is arranged as the y axis in the longitudinal direction of the body. By arranging the vertical direction as the z axis, a triaxial acceleration sensor is configured. That is, the two acceleration ICs are arranged at an angle of 90 degrees with the z-axis in common. Specifically, the first acceleration IC15-1 is placed face down, and the second acceleration IC15-2 is placed upright.

  FIG. 5 is a functional block diagram showing the configuration of the sensor unit 14. The sensor unit 14 includes an acceleration sensor 15, a signal detection circuit 32, a calculation unit (CPU) 30, a random access memory (RAM) 42 that gives a memory area necessary for calculation processing of the calculation unit 30, and a program executed by the calculation unit 30 And a rewritable read-only memory (FEPROM, EEPROM) 44 for storing data. Further, the sensor unit 14 includes a communication module 13 that communicates with a communication device 16 such as a mobile phone as necessary. In this embodiment, the communication module 13 is Bluetooth, but communication means using infrared rays or FM waves can be used instead. As the communication device 16, a wireless terminal device such as a portable information terminal (such as a PDA) can be used instead of a mobile phone. The communication device 16 communicates with the center 70 via the network 11.

  The sensor unit 14 includes a battery 17. The battery 17 may be rechargeable. In this case, the battery 17 is charged by a charger connected to a home or office AC power source.

  The sensor unit 14 processes a signal from the acceleration sensor, stores it in a memory, and monitors body movements by a processing program. If determined to be abnormal, the communication control unit activates Bluetooth and sends acceleration data to the mobile phone 16, and further sends this data to the center 70 via the network 11 such as the Internet.

  The output from the acceleration sensor 15 is A / D converted at a sampling frequency of 100 Hz by the signal detection block 32 of the calculation unit 30, subjected to noise removal processing by the noise removal unit 33, and affine transformed by the 3D affine conversion unit 35. . Processing up to affine transformation is called initialization processing.

  FIG. 6 is a flowchart 51 showing the initialization process of the acceleration sensor. In this initialization process, the initial state when the user wears the monitoring device is assumed to be standing or sitting (401). Acquire acceleration signals Axr, Ayr, and Azr in three directions based on the output of the acceleration sensor 15 (403), pass through a low-pass filter (405), perform decimation / interpolation processing (407), and install a direct current component Axr_0 , Ayr_0 and Azr_0 are acquired (409). Theoretically, coordinate conversion in an arbitrary three-dimensional space has six degrees of freedom. However, if the problem is limited to the rotation of three axes (x, y, z) and there is no movement of the origin, there are three degrees of freedom. (X-axis rotation angle α, y-axis rotation angle β, z-axis rotation angle γ) (FIG. 2).

  α, β, and γ are obtained (411), and an affine transformation matrix is constructed (413). This coordinate transformation is called affine transformation. The specific calculation method and symbol / variable definitions are described below.

The reference coordinate system Cs is defined as follows. (See Figure 1)
1. x-axis = right (+) left (-)
2. y-axis = front (+) back (-)
3. z-axis = up (+) down (-)
The acceleration signal for each axis is defined as follows.

  In the actual coordinate system Cr, variables describing the signals of the x, y, and z axes are Axr, Ayr, and Azr.

  In the reference coordinate system Cs, variables describing the signals of the x, y, and z axes are Axs, Ays, and Azs.

The data conversion matrix from the actual measurement value to the reference coordinate system is Mxyz. Moreover, the x-axis around the rotation matrix M _{X [alpha,} the _M yβ y-axis of the rotation matrix, the z axis of the rotation matrix and _M zγ.

  The rotation angle around each axis in the affine transformation is obtained as follows.

Rotation angle around x-axis α = arctg (Ayr0 / Azr0)
Rotation angle around y-axis β = arctg (Axr0 / Azr0)
Rotation angle around the z axis γ = arctg (Ayr0 / Axr0)
However, Axr0, Ayr0, and Azr0 are DC components whose gravity vectors reflect their projections on the three orthogonal measurement axes when actually mounted. Further, the clockwise direction toward the rotation axis direction is the positive direction of the angle.

A rotation matrix around the x-axis (on the YZ plane) is obtained by the following equation.

A rotation matrix around the y-axis (on the XZ plane) is obtained by the following equation.

A rotation matrix around the z-axis (on the XY plane) is obtained by the following equation.

The transformation matrix Mxyz is obtained from M _xα , M _yβ , and M _zγ by the following formula.
Mxyz = M _xα × M _yβ × M _zγ
In order to perform data conversion from the actual coordinate system to the reference coordinate system using the conversion matrix Mxyz obtained in the initialization process, the following variables are defined.

Matrix Pr = (Axr, Ayr, Azr, 1) ^T before conversion composed of actual measured values
A matrix Ps = (Axs, Ays, Azs, 1) ^T after transformation composed of transformation data to the reference coordinate system
The superscript T indicates a matrix transposition operation.

  Coordinate conversion is performed by the following arithmetic expression.

Ps = Mxyz × Pr = M _xα × M _yβ × M _zγ × Pr
In the above initialization process, a transformation matrix is obtained only at the time of mounting. As long as it is not put on again, the above conversion relationship is effective, and the conversion matrix is used for the following detection of falls, postures, and walking rhythms.

FIG. 7 is a flowchart showing an analysis method of fall detection, posture / posture detection, and walking rhythm detection based on acceleration signals. Signals Axr, Ayr, Azr in the actual measurement coordinate system are obtained from the triaxial acceleration sensor 15 from the signal detection circuit 32 (501), Pr = (Axr, Ayr, Azr, 1) ^T Then, affine transformation is performed using the transformation matrix Mxyz obtained by initialization to obtain Ps (502). From the matrix Ps converted to the reference coordinate system, analyze data using As = (Axs, Ays, Azs), fall (518, 522, 526, 530), posture (538), walking rhythm (549), etc. Make a decision.

  First, in order to remove high-frequency components including noise, the converted acceleration data is passed through a low-pass filter (503) (the cutoff frequency of the low-pass filter is 2 Hz), and the signal DC is obtained by thinning and interpolation (505) processing. The ingredients are removed (506). According to the sampling theory, if sampling is performed at a low frequency (corresponding to thinning), high-frequency components cannot be reproduced and are dropped. Also, when returning to the original 100 Hz sampling frequency by Cubic interpolation, the high frequency component was dropped, so only the smooth baseline (DC component) remains. On the other hand, the output of the step 505 (DC component As_) is subtracted from the output of the low-pass filter 503 (510) to extract the fluctuation component As ~ of the acceleration signal (511).

  Differences ΔAx_, ΔAy_, ΔAz_ are obtained from the DC components Axs_, Ays_, Azs_ of the x-axis, y-axis, and z-axis extracted by the thinning / interpolation process (505) (507).

  The generated difference values ΔAx_, ΔAy_, ΔAz_ are compared with predetermined threshold values Thx (= 0.25G), Thy (= 0.25G), Thz (= 0.25G), respectively, and according to the result A forward fall (517), a backward fall (521), a left fall (525), and a right fall (529) are determined. That is, when -ΔAz_ is larger than the threshold value Thz, the fall is determined, and the direction is determined according to the other difference values ΔAx_ and ΔAy_. When ΔAx_ is smaller than Thx and −ΔAy_ is larger than Thy (517), it is determined that the fall is forward. Similarly, in accordance with the conditions shown in blocks 521, 525, and 529, backward fall, left fall, and right fall are determined.

  FIG. 8 shows the acceleration waveform of each axis when the vehicle falls forward, FIG. 9 shows the acceleration waveform of each axis when the vehicle falls backward, and FIG. 10 shows the acceleration waveform of each axis when the vehicle falls left. 11 shows the acceleration waveform of each axis at the time of falling to the right.

  Fluctuation components Axs˜, Ays˜, Azs˜ of the acceleration signal extracted by the adding unit 510 indicate body fluctuations in the respective directions of the three axes, that is, fluctuation components of acceleration. Therefore, this component is normalized (533), yaw, pitch, and roll angle fluctuations are calculated (535), and through smoothing processing (537), the user's three-dimensional space in the x-axis direction, y-axis direction, and z-axis direction It is possible to detect the posture and movement in For example, in the posture of lying down, the component in the z-axis direction is almost zero, the component in the y-axis direction is maximized, and the component in the x-axis direction does not change. When lying on the back, the component in the z-axis direction is almost zero, the component in the y-axis direction is a negative maximum, that is, the minimum, and the component in the x-axis direction does not change.

  Further, the walking rhythm of the user can be detected using the fluctuation component. The vertical z-axis acceleration component Azs ~ is extracted (541), smoothed by a low-pass filter (543), centered so that the average value of the signal becomes zero (545), and the zero cross method (acceleration waveform and The walking rhythm can be detected by detecting the intersection with the zero baseline.

  FIG. 12 shows an example of walking rhythm detection by the zero cross method. FIG. 12 (a) shows an acceleration waveform in the z-axis direction collected during walking for 10 seconds. FIG. 12 (b) shows the result of processing through a low-pass filter. FIG. 12 (c) shows the result of centering and detecting the walking rhythm by the zero cross method. A small circle mark indicates the detection point of the walking step.

  As is clear from the above description, by performing real-time analysis of the triaxial acceleration signal and the gravity vector, the user's fall can be detected and the number of steps, walking speed, and walking distance can be measured. In addition, it is possible to determine the state of lying down (down), the state of sleeping on the back (back), the state of sleeping sideways, the sitting state (sitting), and the standing state (standing).

  Furthermore, the calorie consumption amount of the user is also obtained from the obtained number of steps. In this calculation, not only the number of steps but also the walking speed can be included in the calculation.

  The present invention collects and analyzes static signals (gravity vectors) and dynamic signals (acceleration data) related to human movement in a three-dimensional space, and as shown in FIG. For example, it can be applied to animal behavior monitoring and robot posture detection, such as detection of falls of elderly people, evaluation of sleep quality, estimation of momentum, detection of drunkness, and the like.

  In addition, the initialization processing for converting the arbitrary sensing coordinate axes to the unified reference coordinate axes eliminates the need for user consideration regarding the mounting direction of the apparatus. Furthermore, as long as you have a mobile phone, you can use it according to your needs as shown in FIG.

  In one form of application, as shown in FIG. 13, a real-time care service is provided not only at home but also for elderly people living in care apartments and nursing homes. The user wears a belt buckle type device. When a user's unexpected fall is detected by real-time analysis of data, information (user ID, time, place, incident) and related data before and after the occurrence point are immediately sent to the computer of the information processing center.

  In addition, if a person who is exercising wears a sensor device, the amount of exercise is estimated by real-time analysis of the data, and if too much exercise is detected, an advice message or alarm is immediately displayed on the user's mobile phone. To do. A system can be constructed that can advise the individual user to adjust to the optimal momentum.

  Furthermore, the analysis of walking rhythm can be applied to the detection of a drunk state after drinking too much. If abnormal turbulence or randomness is detected from the walking rhythm, the possibility of getting drunk can be estimated.

  As mentioned above, although the typical concrete Example of this invention was described, this invention is not limited to such an Example.

The area where the sensor can be mounted. Sensor coordinate system definition. Arrangement of acceleration IC and conceptual diagram of sensor unit. The conceptual diagram which shows various mounting methods. The functional block diagram of a sensor unit. The flowchart which shows the flow of the initialization process of a sensor. The flowchart which shows the flow of the process which detects a user's attitude | position, a fall, and a walking rhythm. The figure which shows the waveform at the time of forward falling. The figure which shows the waveform at the time of falling backward. The figure which shows the waveform at the time of falling to the left. The figure which shows the waveform at the time of falling to right. An example of walking rhythm detection by the zero cross method. The figure which shows the example of application of this invention.

Explanation of symbols

14 Sensor unit 15 Acceleration IC

Claims (6)

An acceleration sensor that detects accelerations of three orthogonal axes and measures each acceleration data;
A sensor unit having a data processing device having an initialization processing function for converting data from the acceleration sensor into a three-axis reference coordinate system;
A body condition monitoring device that is attached to a user's body and collects information related to the movement of the user.   The physical condition monitoring device according to claim 1, wherein the sensor unit is configured to detect a user's fall based on a direct current component of a vertical axis (z-axis) signal from the triaxial acceleration sensor.   The sensor unit is configured to detect a user's fall direction based on a DC component of a signal of a front-rear axis (y-axis) or a left-right axis (x-axis) from the triaxial acceleration sensor. The physical condition monitoring apparatus according to claim 2.   The physical condition monitoring apparatus according to claim 2, wherein the sensor unit is configured to detect a user's fall when a difference in moving average value of DC components of the vertical axis signal is greater than a predetermined value.   When the difference of the moving average value of the DC component of the signal on the front-rear axis (y-axis) or the left-right axis (x-axis) from the three-axis acceleration sensor is greater than a predetermined value, the sensor unit The body condition monitoring apparatus according to claim 4, wherein the body condition monitoring apparatus is configured to detect a direction of falling (front / rear / left / right).   The physical condition monitoring apparatus according to claim 1, wherein the sensor unit is configured to detect a user's posture based on a moving average value of a fluctuation component of each of three-axis signals from the three-axis acceleration sensor.

Images